A ‘data point’ was defined as a pre- or post-introduction prevalence in a single year, age group, and population. A ‘data set’ was

defined as two data points, separated in time, from the same age group and population, typically one pre- and one post- introduction. Where possible, the ‘pre’ period was before PCV licensing in the country, excluding the year licensed unless that year’s pre-data were drawn only from months prior to introduction (Appendix B.1); the ‘post’ period began no earlier than the year following introduction. ZD1839 Year of introduction was based on a compilation of data from WHO [19] and VIMS [20] databases which identified the year in which PCV was widely adopted on a national or relevant regional scale. In the few cases with significant lag time between national licensure and wide adoption, the breakpoint identified by the author was used (low-coverage vs. high-coverage, or pre-licensure vs. post-licensure.) Percentage change in outcome measures was calculated by comparing the most recent pre-introduction data available to each available post-introduction time point. For data presented as incidence rates and case counts, percentage change was calculated as

(pre-introduction – post-introduction)/pre-introduction × 100%, where negative XAV-939 mw values for percentage change denote an increase. If the study outcome was the proportion VT of all IPD cases, percentage change was transformed into a comparable measure based on incidence rates and case counts as follows: Percentage change = [1 − ((%VT IPD post) × (%NVT IPD pre))/(%VT IPD pre) × (%NVT IPD post)] × 100%. Data were stratified by elapsed years since introduction to assess trends with time, and by age group (<5, 5 to <18, 18 to <50, 50 to <65, ≥65 years) to assess differential effects across age categories. Points not fitting within a single age stratum with minimal overlap

were classified based on the oldest stratum Libraries included. Where a data point represented multiple post-introduction out years (i.e., “2001–2003”), the midpoint was used to calculate the number of years since PCV introduction. Where possible, data were also stratified into populations receiving booster doses and those without, and indigenous versus general populations. Effects of different primary dose schedules are addressed elsewhere [21], [22], [23] and [24]. When both IPD and carriage were available, we compared their percentage changes to assess their relationship. When both VT-IPD and PCV coverage levels in the community over time were available, we evaluated the relationship between PCV uptake and VT-IPD impact. Countries that implemented a catch-up schedule in those <2 or <5 years were identified; since catch-up coverage is generally less than complete, we did not further distinguish the magnitude of indirect effects by use of catch-up but considered these mixed populations.